The EU AI-Act in Focus
EU Commission publishes Commission Guidelines on prohibited artificial intelligence practices
The European Commission has published official guidelines on the implementation of the prohibitions on certain AI practices set out in the EU AI Act. The aim of the guidelines is to support companies, authorities and other stakeholders in the EU in applying the new regulations in a uniform and legally compliant manner.
Background to the AI-Act
The EU AI Act (Regulation (EU) 2024/1689) is a set of rules for the regulation of artificial intelligence that takes a risk-based approach. While many AI systems may continue to be used freely, high-risk applications are subject to strict conditions, while some technologies with unacceptable risk are completely banned under Article 5 of the regulation. These bans have now come into force on February 2, 2025.
Significance of the new guidelines
The 140-page guidelines published by the European Commission contain detailed explanations of the prohibited practices. The guidelines provide important guidance for companies that need to adapt to the new regulations.
The Commission emphasizes that the prohibitions and their implementation are essential for safeguarding European fundamental rights. In particular, the distinction between prohibited and high-risk AI systems poses a challenge for companies, which the guidelines are intended to mitigate.
While the guidelines are not legally binding, they provide the EU Commission’s interpretation of prohibited AI practices and are intended to ensure uniform implementation of the AI-Act in all member states. However, the final legal interpretation is left to the courts.
What is prohibited?
The in the EU AI Act prohibited practices include:
- Manipulative or deceptive AI systems that unconsciously influence the behavior of users
- Social scoring by authorities that rates citizens based on their behavior
- Mass biometric surveillance, especially in real time
- Recognition of emotions in work or educational institutions to evaluate people’s feelings
- AI-supported crime predictions that predict the potential risk of a crime based on profiling or personal characteristics
- The use of AI to create so-called “dark patterns” that tempt or manipulate users into certain actions
- Chatbots with subliminal manipulation, which encourage people to make harmful financial decisions
Impact on companies and authorities
As the bans apply immediately, companies must review and adapt their AI products and services. Public authorities using AI systems will also need to ensure that they are not using prohibited technologies. While the guidelines are not legally binding, they serve as a benchmark for enforcement by supervisory authorities.
The Commission emphasizes that violations against the EU AI Act can be punished with severe penalties – up to 35 million euros or 7% of a company’s annual global turnover.
The next stage of AI-Act obligations, which focuses on general-purpose AI models, will take effect in August 2025. All provisions of the AI-Act must be fully implemented by August 2026.
Conclusion
By publishing the guidelines, the EU is creating more legal certainty for companies and public bodies. As the AI Act must be fully implemented by August 2026, all affected stakeholders should prepare for this at an early stage to avoid penalties and business risks. The guidelines are a key part of this transition and provide important guidance for the entire industry.